Prognostic outcomes of adjuvant chemotherapy in postoperative patients with cancer undergoing hemodialysis versus non-hemodialysis patients: a retrospective nationwide cohort study Authors: Taisuke Ishii, Tomone Watanabe, Yuichi Ichinose, Hiroyuki Mano, Takahiro Higashi Published: 2026 ABSTRACT AbstractBackgroundSurvival outcomes for patients with cancer undergoing hemodialysis (HD) are poorer than those for patients not undergoing HD, possibly owing to altered chemotherapy protocols or undertreatment to address renal impairment and drug toxicity concerns. Adjuvant chemotherapy should be administered appropriately with evidence-based drug selection. However, in patients undergoing HD, the administration of adjuvant chemotherapy and its association with prognosis remain unclear. We evaluated the association between adjuvant chemotherapy and survival outcomes in patients undergoing HD compared with those not undergoing HD in Japan.MethodsA nationwide dataset combining Hospital-Based Cancer Registry and Diagnosis Procedure Combination survey data was utilized. The study included adult patients newly diagnosed with colorectal, gastric, breast, or non-small cell lung cancer between January 2013 and December 2015. The primary survival analysis used a 90-day landmark approach: patients who were alive at postoperative day 90 and had survival follow-up information available thereafter were included, adjuvant chemotherapy was defined as treatment initiated within 90 days after surgery, and overall survival was measured from postoperative day 90 to death or last confirmed survival. Multivariable Cox regression models included hemodialysis status, adjuvant chemotherapy, their interaction term, age, sex, cancer type, Barthel Index, comorbidities, and hospital type.ResultsIn the final cohort, 44,281 patients met the eligibility criteria. Among 43,984 patients included in the primary 90-day landmark analysis, 451 were undergoing HD. Patients undergoing HD were less likely to initiate adjuvant chemotherapy within 90 days after surgery than those not undergoing HD (30.4% vs. 68.0%, P < 0.001). In the multivariable Cox model with an interaction term, HD was associated with worse overall survival among patients who did not initiate adjuvant chemotherapy within 90 days after surgery (adjusted hazard ratio [aHR], 1.91; 95% confidence interval [CI], 1.67–2.18; P < 0.001), and the HD × adjuvant chemotherapy interaction was statistically significant (P = 0.002). The association between adjuvant chemotherapy and overall survival was evident among patients not undergoing HD but was not statistically significant among patients undergoing HD. In an exploratory analysis restricted to patients undergoing HD who initiated adjuvant chemotherapy, use of a standard regimen was associated with better survival than use of a non-standard regimen.ConclusionsPatients undergoing HD were less likely to receive adjuvant chemotherapy and had worse overall survival. The association between adjuvant chemotherapy and survival appeared attenuated among patients undergoing HD. Further prospective studies are needed to define optimal postoperative chemotherapy strategies for this population.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12885-026-16381-9.Keywords: Hemodialysis, Adjuvant chemotherapy, Cancer, Chemotherapy regimens, Survival Background Patients undergoing chronic hemodialysis (HD) face a significantly higher risk of cancer than the general population, with reported cancer incidence rates up to 1.44 times higher and cancer mortality rates 1.5–2.9 times higher than in individuals not undergoing HD [1–4]. Advances in dialysis have extended the lifespan of patients undergoing HD [5, 6] but also increased their likelihood of developing cancer. Evidence-based standard care should, therefore, be developed specifically for this population. Adjuvant chemotherapy has been shown to improve survival in various cancers, including colorectal, gastric, non-small cell lung cancer (NSCLC), and breast cancer [7–10]. However, clinical trials establishing these benefits have largely excluded patients undergoing HD, leaving significant gaps in evidence for this group. Previous studies have indicated that patients undergoing HD often receive modified or reduced doses of adjuvant chemotherapy due to concerns about increased toxicity, altered drug clearance, and greater sensitivity to these treatments [11, 12]. The long-term survival impact of these modifications remains poorly understood. Moreover, common comorbidities—such as cardiovascular disease and diabetes—and reduced physical fitness, further complicate cancer treatment in patients undergoing HD, potentially compromising outcomes [13, 14]. The lack of reliable data on adjuvant chemotherapy administration and its impact on patients undergoing HD poses significant challenges for clinical decision-making. Oncologists require tailored treatment protocols that ensure both the efficacy and safety of cancer treatment for this group [15–17]. Without evidence-based guidance, treatment decisions often rely on clinical judgment, resulting in potentially suboptimal care for this high-risk population [15]. The aim of this study was to address current knowledge gaps by evaluating the association between adjuvant chemotherapy and survival outcomes in patients with cancer, comparing those undergoing HD with patients not undergoing HD using a landmark approach to reduce immortal time bias and accounting for key prognostic factors. These findings are intended to inform the development of individualized treatment approaches for patients with cancer requiring HD. Methods Data source and study approval We utilized a comprehensive dataset linking the Hospital-Based Cancer Registry (HBCR) with Diagnosis Procedure Combination (DPC) survey data, enabling analysis of healthcare utilization, survival outcomes, and clinical characteristics. The HBCR, a mandatory cancer-reporting system for hospitals designated by the Ministry of Health, Labour and Welfare in Japan, also includes voluntary participation from non-designated hospitals serving similar roles in their communities [18, 19]. It records basic clinical data such as age, sex, cancer type, stage, tumor-node-metastasis classification, histopathology (ICD-O-3), and the treating hospital. Additionally, the HBCR tracks 5-year survival by checking resident records through inquiries to local government offices. DPC survey data provides details on all billable health services, including surgeries, prescriptions, and hemodialysis, for both outpatient and inpatient care. Because DPC data are submitted on a Japanese fiscal-year basis (April through March), we obtained DPC records for HBCR-registered patients covering the period from the October preceding diagnosis through March of the second fiscal year after diagnosis, capturing perioperative care and subsequent treatments for 65.4% of HBCR-registered patients. This linkage enabled the analysis of cancer-related information, treatments, and 5-year prognosis. The Barthel Index [20] (assessing activities of daily living) and comorbidities (e.g., cardiovascular diseases) were identified using DPC data. Comorbidities were defined based on the Charlson Comorbidity Index using ICD-10 codes registered in the DPC survey data [21, 22]. The Institutional Review Board of the National Cancer Center, Japan, approved this study (approval number 2022 − 139). This study followed the principles outlined in the Declaration of Helsinki. This study was an observational study using existing data, and participants provided informed consent using an opt-out method, in accordance with the Act on the Protection of Personal Information and ethical guidelines in Japan. No one declined to participate in the study. Patient sample The study analyzed patients aged ≥ 20 years newly diagnosed with colorectal, gastric, breast, or NSCLC between January 2013 and December 2015, as postoperative adjuvant chemotherapy for these cancer types is well established. We focused on patients in stages where adjuvant chemotherapy is commonly recommended: pStage II/III for gastric cancer (excluding pT1 and pT3N0), pStage III for colorectal cancer, pStage IIB or IIIA for NSCLC (excluding those receiving chemoradiotherapy), and pStage III for breast cancer (excluding pT3N1). We limited the breast cancer cohort to Stage III disease because adjuvant chemotherapy is generally considered standard in this group; although patients with Stage I–II breast cancer may receive adjuvant chemotherapy depending on tumor biology and recurrence risk, such indications could not be reliably ascertained using administrative claims data. After exclusions (526 patients with neoadjuvant chemotherapy, 1,419 lacking a Barthel Index at discharge, and 795 without hospital follow-up), the final cohort included 44,281 patients from 400 facilities (75.0% being designated cancer centers). The study achieved a survival status capture rate of 95.8%. Definitions Adjuvant chemotherapy was defined as chemotherapy initiated within 90 days after surgery for all cancer types in the primary analysis. In the 90-day landmark analysis, overall survival (OS) was calculated from postoperative day 90 to either the date of death or last confirmed survival. HD status was identified using DPC codes linked to HD services. Contraindicated chemotherapy agents for patients undergoing HD included specific drugs such as tegafur-gimeracil-oteracil (S-1), oxaliplatin, cisplatin, and pemetrexed, which are components of standard regimens such as folinic acid plus 5-fluorouracil plus oxaliplatin and S-1 plus oxaliplatin, as outlined in the Japanese guidelines [23]. Statistical analysis For bivariate analysis, we utilized the Student’s t-test or Mann–Whitney U test for continuous variables, depending on data distribution, and used the chi-squared test for categorical variables. For the primary survival analysis, we used a 90-day landmark approach to reduce immortal time bias. Patients who were alive at postoperative day 90 and had survival follow-up information available thereafter were included in the 90-day landmark cohort. Adjuvant chemotherapy exposure was defined as treatment initiation within 90 days after surgery, and survival time was measured from postoperative day 90. To evaluate whether the association between adjuvant chemotherapy and survival differed by HD status, we used multivariable Cox regression analysis including the interaction term between HD status and adjuvant chemotherapy. Covariates included age, sex, cancer type, Barthel Index, comorbidities, and hospital type. Cancer type, rather than pathological stage, was included in the primary model because the stage distribution was structurally linked to cancer type by the eligibility criteria; all colorectal and breast cancer patients had Stage III disease, resulting in sparse or empty stage strata for some cancer types. We also performed subgroup analyses to examine whether the association between adjuvant chemotherapy and OS was consistent across clinically relevant patient and hospital characteristics, including HD status, age, sex, Barthel Index, comorbidity status, and hospital type. These subgroup models were adjusted for the same covariates as the primary model, except for the stratifying variable. In the analysis restricted to patients undergoing HD who initiated adjuvant chemotherapy at any time after surgery, OS was measured from the date of adjuvant chemotherapy initiation. Because this subgroup included only 147 patients and some cancer-type strata were very small, cancer type and pathological stage were not included as covariates in this exploratory model to avoid unstable estimates. Sensitivity analyses were performed using two additional landmark approaches. First, we repeated the 90-day landmark analysis among patients who were discharged within 31 days after surgery. In this analysis, adjuvant chemotherapy exposure was defined as treatment initiation within 90 days after surgery, and survival time was measured from postoperative day 90. Second, we performed a 45-day landmark analysis, including patients who were alive at postoperative day 45 and had survival follow-up information available thereafter. In this analysis, adjuvant chemotherapy exposure was defined as treatment initiation within 45 days after surgery, and survival time was measured from postoperative day 45. As an additional sensitivity analysis, we repeated the primary multivariable Cox model with pathological stage substituted for cancer type. All statistical tests were two-sided and performed using Stata version 17.0 (StataCorp LP, College Station, TX, USA). Bar graphs were created with GraphPad Prism 9 (GraphPad Software Inc., San Diego, CA, USA). A P-value of < 0.05 was considered statistically significant. Results Patient characteristics Figure 1 shows patient selection and the analytic populations. The final cohort included 44,281 patients, and the primary 90-day landmark analysis included 43,984 patients who were alive at postoperative day 90 and had survival follow-up information available thereafter. Table 1 provides the baseline characteristics of these patients stratified by HD status. Among the 43,984 patients in the primary analysis, 451 were undergoing HD. Patients undergoing HD were less likely to initiate adjuvant chemotherapy within 90 days after surgery than patients not undergoing HD (30.4% vs. 68.0%, P < 0.001). The Barthel Index revealed reduced functional independence in the HD group, as 78.0% achieved a score of 100 compared with 91.7% in the non-HD group (P < 0.001). Patients who died during follow-up were older, had lower activity levels, and had a higher prevalence of comorbidities than those who were alive (Additional file 1, Table 1). Patient selection. We identified patients aged ≥ 20 years who were newly diagnosed between 2013 and 2015 with gastric, colorectal, non-small cell lung, or breast cancer, and had postoperative pathological stages eligible for adjuvant chemotherapy (N = 47,021). We excluded 526 patients who received preoperative chemotherapy, 1,419 patients without a documented Barthel Index, and 795 patients who did not have any follow-up visits at the same facility after surgery. The final cohort included 44,281 patients. The primary 90-day landmark analysis and sensitivity analysis 3 included 43,984 patients who were alive at postoperative day 90 and had survival follow-up information available thereafter. Sensitivity analysis 1 included 40,829 patients who were discharged within 31 days after surgery, were alive at postoperative day 90, and had survival follow-up information available thereafter. Sensitivity analysis 2 included 44,178 patients who were alive at postoperative day 45 and had survival follow-up information available thereafter Table 1Patient characteristics stratified according to hemodialysis status in the 90-day landmark cohortCharacteristicsTotal (n = 43,984)Non-dialysis (n = 43,533)Dialysis (n = 451)P-valueAge in years, median (IQR)70 (62–77)70 (62–77)70 (64–78)0.007 <65, n (%)14,132 (32.1)14,015 (32.2)117 (25.9)0.015 65–75, n (%)15,850 (36.0)15,678 (36.0)172 (38.1) ≥75, n (%)14,002 (31.8)13,840 (31.8)162 (35.9)Sex (female), n (%)19,356 (44.0)19,246 (44.2)110 (24.4)< 0.001Cancer type, n (%)< 0.001 Gastric cancer12,938 (29.4)12,799 (29.4)139 (30.8) Colorectal cancer22,428 (51.0)22,164 (50.9)264 (58.5) NSCLC5,074 (11.5)5,050 (11.6)24 (5.3) Breast cancer3,544 (8.1)3,520 (8.1)24 (5.3)Pathological stage, n (%)0.15 Stage II7,448 (16.9)7,383 (17.0)65 (14.4) Stage III36,536 (83.1)36,150 (83.0)386 (85.6)Postoperative length of stay, median (IQR; days)12 (9–16)12 (9–16)17 (11–31)< 0.001Adjuvant chemotherapy initiated within 90 days after surgery, n (%)29,750 (67.6)29,613 (68.0)137 (30.4)< 0.001Standard regimen initiated within 90 days after surgery, n (%)26,187 (59.5)26,085 (59.9)102 (22.6)< 0.001Drugs contraindicated for patients undergoing HD initiated within 90 days after surgery, n (%)19,156 (43.6)19,104 (43.9)52 (11.5)< 0.001Barthel Index< 0.001 10040,295 (91.6)39,943 (91.7)352 (78.0) 85–952,516 (5.7)2,449 (5.6)67 (14.9) <851,173 (2.7)1,141 (2.6)32 (7.1)Comorbidity, n (%)12,885 (29.3)12,744 (29.3)141 (31.3)0.36Hospital type, n (%)0.72 Non-designated hospital4,114 (9.4)4,074 (9.4)40 (8.9) Designated hospital39,870 (90.7)39,459 (90.6)411 (91.1)Year of diagnosis, n (%)0.93 201311,160 (25.4)11,044 (25.4)116 (25.7) 201414,680 (33.4)14,527 (33.4)153 (33.9) 201518,144 (41.3)17,962 (41.3)182 (40.4)Abbreviations: IQR Interquartile range, NSCLC Non-small-cell lung cancer, HD Hemodialysis Patient characteristics stratified according to hemodialysis status in the 90-day landmark cohort Abbreviations: IQR Interquartile range, NSCLC Non-small-cell lung cancer, HD Hemodialysis Survival analysis Table 2 shows the factors associated with OS in the primary 90-day landmark analysis. The interaction term between HD and adjuvant chemotherapy initiated within 90 days after surgery was statistically significant (P = 0.002), indicating that the association between adjuvant chemotherapy and OS differed by HD status. In this interaction model, HD was associated with worse OS among patients who did not initiate adjuvant chemotherapy within 90 days after surgery (adjusted HR, 1.91; 95% CI, 1.67–2.18; P < 0.001). Adjuvant chemotherapy was associated with better OS among patients not undergoing HD (adjusted HR, 0.61; 95% CI, 0.59–0.64; P < 0.001), whereas this association was not statistically significant among patients undergoing HD (adjusted HR, 0.93; 95% CI, 0.72–1.21; P = 0.61). Table 2Factors associated with overall survival in the 90-day landmark cohort (N = 43,984)CharacteristicsUnadjusted hazard ratio (95% CI)P-valueAdjusted hazard ratio (95% CI)aP-valueHemodialysis2.63 (2.35–2.95)< 0.0011.91 (1.67–2.18)< 0.001Adjuvant chemotherapy initiated within 90 days after surgery0.47 (0.45–0.48)< 0.0010.61 (0.59–0.64)< 0.001Hemodialysis × Adjuvant chemotherapy initiated within 90 days after surgery (interaction)——1.52 (1.17–1.98)0.002Age (years) <65ReferenceReference 65–751.47 (1.41–1.54)< 0.0011.26 (1.21–1.32)< 0.001 ≥752.74 (2.63–2.85)< 0.0011.95 (1.86–2.04)< 0.001Sex MaleReferenceReference Female0.68 (0.66–0.70)< 0.0010.79 (0.76–0.81)< 0.001Cancer type Gastric cancerReferenceReference Colorectal cancer0.48 (0.47–0.50)< 0.0010.50 (0.48–0.51)< 0.001 NSCLC0.79 (0.76–0.83)< 0.0011.03 (0.98–1.08)0.21 Breast cancer0.33 (0.31–0.36)< 0.0010.42 (0.39–0.46)< 0.001Barthel Index 100ReferenceReference 85–952.18 (2.07–2.31)< 0.0011.57 (1.49–1.67)< 0.001 <853.25 (3.03–3.49)< 0.0012.29 (2.13–2.46)< 0.001Comorbidity NoReferenceReference Yes1.42 (1.37–1.47)< 0.0011.14 (1.10–1.18)< 0.001Hospital type Non-designatedReferenceReference Designated0.99 (0.94–1.05)0.711.00 (0.94–1.05)0.86Abbreviations: CI Confidence interval, NSCLC Non-small-cell lung canceraAdjusted for hemodialysis, adjuvant chemotherapy, interaction term of hemodialysis and adjuvant chemotherapy, age at diagnosis, sex, cancer type, Barthel Index, comorbidity, and hospital type Factors associated with overall survival in the 90-day landmark cohort (N = 43,984) Abbreviations: CI Confidence interval, NSCLC Non-small-cell lung cancer aAdjusted for hemodialysis, adjuvant chemotherapy, interaction term of hemodialysis and adjuvant chemotherapy, age at diagnosis, sex, cancer type, Barthel Index, comorbidity, and hospital type Subgroup analysis of adjuvant chemotherapy and OS Figure 2 illustrates the association between adjuvant chemotherapy and OS across selected patient and hospital subgroups in the primary 90-day landmark analysis. Among patients not undergoing HD, adjuvant chemotherapy was associated with better OS, whereas this association was not statistically significant among patients undergoing HD. The favorable association of adjuvant chemotherapy was generally consistent across age categories, sex, Barthel Index categories, comorbidity status, and hospital type. Association between adjuvant chemotherapy and overall survival across selected patient and hospital subgroups in the 90-day landmark cohort (N = 43,984). Adjuvant chemotherapy was associated with better overall survival among patients not undergoing hemodialysis, whereas this association was not statistically significant among patients undergoing hemodialysis. The favorable association of adjuvant chemotherapy was generally consistent across age categories, sex, Barthel Index categories, comorbidity status, and hospital type Survival in patients undergoing HD receiving adjuvant chemotherapy Table 3 focuses on the survival outcomes of 147 patients undergoing HD who initiated adjuvant chemotherapy at any time during follow-up. In this exploratory subgroup analysis, OS was defined as the duration from the day of adjuvant chemotherapy initiation to the date of last follow-up. The use of a standard chemotherapy regimen was associated with better survival (adjusted HR, 0.48; 95% CI, 0.30–0.77; P = 0.003). Table 3Factors associated with overall survival among all postoperative patients undergoing hemodialysis who received adjuvant chemotherapy at any time after surgery (N = 147)CharacteristicsUnadjusted hazard ratio (95% CI)P-valueAdjusted hazard ratio (95% CI)aP-valueStandard regimen0.53 (0.34–0.84)0.0070.48 (0.30–0.77)0.003Age [years] < 65ReferenceReference 65–751.33 (0.82–2.17)0.251.36 (0.83–2.24)0.23 ≥ 751.86 (1.02–3.37)0.042.21 (1.17–4.14)0.01Sex MaleReferenceReference Female0.66 (0.39–1.12)0.130.69 (0.40–1.17)0.17Barthel Index 100ReferenceReference 85–950.98 (0.40–2.42)0.970.88 (0.35–2.21)0.79 < 851.57 (0.57–4.30)0.381.42 (0.50–4.07)0.50Comorbidity NoReferenceReference Yes1.27 (0.81–1.99)0.301.14 (0.72–1.82)0.58Hospital type Non-designatedReferenceReference Designated1.58 (0.69–3.64)0.281.15 (0.49–2.71)0.74Abbreviations: CI Confidence intervalaAdjusted for standard chemotherapy regimen, age at diagnosis, sex, Barthel Index, comorbidities, and hospital type Factors associated with overall survival among all postoperative patients undergoing hemodialysis who received adjuvant chemotherapy at any time after surgery (N = 147) Abbreviations: CI Confidence interval aAdjusted for standard chemotherapy regimen, age at diagnosis, sex, Barthel Index, comorbidities, and hospital type Sensitivity analysis Sensitivity analyses showed results consistent with the primary analysis. In the 90-day landmark analysis restricted to patients discharged within 31 days after surgery, the analysis included 40,829 patients who were alive at postoperative day 90 and had survival follow-up information available thereafter. HD was associated with worse OS (adjusted HR, 2.20; 95% CI, 1.93–2.52; P < 0.001), adjuvant chemotherapy initiated within 90 days after surgery was associated with better OS (adjusted HR, 0.62; 95% CI, 0.60–0.65; P < 0.001) among patients not undergoing HD, and the interaction between HD and adjuvant chemotherapy remained statistically significant (P = 0.003) (Additional file 1, Table 2). In the 45-day landmark analysis, the analysis included 44,178 patients who were alive at postoperative day 45 and had survival follow-up information available thereafter. HD was associated with worse OS (adjusted HR, 2.06; 95% CI, 1.82–2.33; P < 0.001), adjuvant chemotherapy initiated within 45 days after surgery was associated with better OS (adjusted HR, 0.69; 95% CI, 0.67–0.72; P < 0.001) among patients not undergoing HD, and the interaction term was statistically significant (P < 0.001) (Additional file 1, Table 3). When pathological stage was substituted for cancer type in the primary multivariable model, the principal findings were unchanged: HD remained associated with worse OS (adjusted HR, 1.90; 95% CI, 1.69–2.13; P < 0.001), adjuvant chemotherapy was associated with better OS (adjusted HR, 0.65; 95% CI, 0.63–0.67; P < 0.001) among patients not undergoing HD, and the HD × adjuvant chemotherapy interaction remained statistically significant (P = 0.01) (Additional file 1, Table 4). Discussion In this nationwide cohort using a 90-day landmark approach, HD was associated with worse OS among postoperative patients with cancer who were eligible for adjuvant chemotherapy. Adjuvant chemotherapy initiated within 90 days after surgery was associated with better OS among patients not undergoing HD, whereas this association was not statistically significant among patients undergoing HD. In an exploratory analysis restricted to patients undergoing HD who initiated adjuvant chemotherapy, use of a standard regimen was associated with better survival than use of a non-standard regimen. HD was associated with a poorer prognosis in the landmark analysis. Patients undergoing HD often have comorbid cardiovascular diseases, which are major contributors to mortality [24, 25]. However, even after adjusting for comorbidities, HD remained associated with poor OS. A possible explanation for the poor prognosis is the higher incidence of postoperative complications among patients undergoing HD, which may negatively affect survival outcomes. Previous studies showed that patients with cancer undergoing HD frequently developed postoperative complications [26–30]. However, sensitivity analyses limited to patients discharged within 31 days postoperatively showed consistent results. Another potential reason for poor prognosis could be the unique nature of cancer in patients undergoing HD. Patients undergoing HD are immunocompromised, which increases their susceptibility to immunologically mediated tumors [3, 31]. Additionally, some studies suggest that tumors in patients with kidney disease may have a distinct immune microenvironment compared with those in patients without kidney disease [32–34]. Further research is warranted to comprehend the biological characteristics of cancer in patients undergoing HD and develop appropriate therapeutic strategies for this population. Our results align with previous studies indicating that adjuvant chemotherapy is associated with improved survival in the general population [35–37]; however, this association was attenuated and not statistically significant among patients undergoing HD. This should not be interpreted as evidence that adjuvant chemotherapy lacks benefit in patients undergoing HD. The attenuated association may reflect residual confounding, competing non-cancer mortality, differences in pharmacokinetics, adverse effects, treatment modification, or cancer biology specific to this population. Emerging evidence shows that uremia and chronic inflammation in patients undergoing HD may contribute to poor responses to therapy [38, 39]. These findings emphasize the need for research aimed at understanding the biological underpinnings of cancer in patients undergoing HD and developing tailored therapeutic strategies. Among patients undergoing HD who initiated adjuvant chemotherapy, standard-regimen use showed a favorable association with survival in exploratory analyses. Our previous study demonstrated that, in patients undergoing HD, concerns regarding the adverse effects of renally excreted chemotherapies often led to a preference for chemotherapies that do not require dose adjustment, even if they are non-standard treatments [16]. The current study similarly showed that non-standard regimens were commonly used, and agents contraindicated for dialysis were less frequently chosen. However, because this subgroup analysis was based on only 147 patients and may be affected by selection bias and residual confounding, this finding should be considered hypothesis-generating. Future prospective studies focusing on specific cancer types are needed to compare standard regimens with modified or less intensive regimens in patients undergoing HD. This study has several limitations. First, because this was a retrospective observational study, the findings should not be interpreted as causal. Although the 90-day landmark approach reduced immortal time bias in the primary analysis, residual selection bias and confounding by indication may remain because patients who initiated adjuvant chemotherapy were likely to have better postoperative recovery, functional status, and clinical profiles than those who did not. We were unable to fully account for performance status, detailed postoperative complications, physician decision-making, patient preferences, and some comorbidities. Second, OS may not fully reflect the oncologic benefit of adjuvant chemotherapy in patients undergoing HD, as non-cancer mortality is substantial in this population. Information on cause of death and recurrence was not available; consequently, cancer-specific survival and competing-risk analyses could not be performed. Nevertheless, the consistent direction of the hazard ratios for HD across the primary, restricted-discharge, and 45-day landmark sensitivity analyses suggests that the worse OS associated with HD is unlikely to be entirely explained by short-term postoperative or non-cancer mortality, although competing non-cancer mortality may still attenuate the apparent benefit of adjuvant chemotherapy in this population. Third, this analysis was limited to facilities that submitted both HBCR and DPC data to our nationwide project, and treatments performed after transfer to other facilities could not be fully captured, which may have led to an overestimation or underestimation of the results. Fourth, the small number of patients undergoing HD, particularly those who initiated adjuvant chemotherapy, limited statistical power and precluded stable adjustment for cancer type in the standard-regimen analysis. Despite these limitations, our nationwide analysis in Japan suggests that postoperative patients with cancer undergoing HD had poorer OS and that the association between adjuvant chemotherapy and OS differed by HD status. Conclusions Among postoperative patients with cancer who were eligible for adjuvant chemotherapy and were alive at postoperative day 90 with survival follow-up information available thereafter, HD was associated with worse OS and a lower likelihood of adjuvant chemotherapy initiation. Adjuvant chemotherapy was associated with better OS among patients not undergoing HD, whereas this association was not statistically significant among patients undergoing HD. Exploratory findings among patients undergoing HD who initiated adjuvant chemotherapy suggest that standard regimens warrant further study. Prospective research is needed to clarify optimal postoperative chemotherapy strategies for patients undergoing HD. Supplementary Information Additional file 1: Table 1. Patient characteristics stratified according to survival status in the 90-day landmark cohort. Table 2. Factors associated with overall survival in the 90-day landmark cohort restricted to patients with a postoperative hospital length of stay ≤31 days (N = 40,829). Table 3. Factor k cohort (N = 44,178)Table 4. Factors associated with overall survival in the 90-day landmark cohort with pathological stage substituted for cancer type (N = 43,984). Acknowledgements We thank all tumor registrars and hospital staff for enabling Hospital-Based Cancer Registries and participating in our quality-of-care project. While writing this manuscript, the authors used ChatGPT (GPT-4o, OpenAI) to enhance readability and proofread the English text. After this, the authors reviewed and edited the content as required and took full responsibility for the content of the publication. We would like to thank Editage (www.editage.jp) for English language editing. Abbreviations Confidence interval Chronic kidney disease Hemodialysis Non-small cell lung cancer Authors’ contributions TI: conceptualization, data curation, formal analysis, project administration, writing—original draft, and writing—review and editing. TW: data curation, methodology, supervision, writing—original draft, and writing—review and editing. YI: data curation and writing—review and editing. HM: supervision and writing—review and editing. TH: formal analysis, methodology, supervision, and writing—review and editing. Funding This work was supported by the Japan Society for the Promotion of Science KAKENHI Grant, funded by the Ministry of Education, Culture, Sports, Science, and Technology of Japan [grant number: 23K16343]. The funder had no role in the conceptualization, design, data collection, analysis, decision to publish, or preparation of the manuscript. Data availability The datasets generated and/or analysed during the current study were used with special permission granted by privacy protection law and data use agreement with the hospitals. The data are not publicly available due to privacy or ethical restrictions. Declarations Ethics approval and consent to participate The Institutional Review Board of the National Cancer Center, Japan, approved this study (approval number 2022 − 139). This study followed the principles outlined in the Declaration of Helsinki. This study was an observational study using existing data, and participants provided informed consent using an opt-out method, in accordance with the Act on the Protection of Personal Information and ethical guidelines in Japan. No one declined to participate in the study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Additional file 1: Table 1. Patient characteristics stratified according to survival status in the 90-day landmark cohort. Table 2. Factors associated with overall survival in the 90-day landmark cohort restricted to patients with a postoperative hospital length of stay ≤31 days (N = 40,829). Table 3. Factor k cohort (N = 44,178)Table 4. Factors associated with overall survival in the 90-day landmark cohort with pathological stage substituted for cancer type (N = 43,984). Data Availability Statement The datasets generated and/or analysed during the current study were used with special permission granted by privacy protection law and data use agreement with the hospitals. The data are not publicly available due to privacy or ethical restrictions.